Salt and pepper noise removal from document images

Abstract

In this paper a noise removal algorithm is proposed by adding a procedure to enhance noise removal to a third party algorithm as a post processing step. The procedure (TAMD) has been proposed to enhance salt and pepper noise removal. TAMD analyzes thin line blobs before deciding to retain or remove them. It has been successfully applied previously in two noise removal algorithms by integrating their algorithm logic with the procedure. In this paper, a noise removal algorithm is proposed by utilizing it as a post processing step. The performance of the proposed noise removal algorithm is compared to many other algorithms including state of the art methods such as median and center weighted median. Real scanned images of mechanical engineering drawings corrupted by 20% salt and pepper noise are used in the experiment. Objective performance evaluation (PSNR and DRDM) has shown that our proposed noise removal algorithm is better than other studied algorithms. © 2009 Springer-Verlag.